Posts in category: nVidia AI

MLPerf remains the definitive measurement for AI performance as an independent, third-party benchmark. NVIDIA’s AI platform has consistently shown leadership across both training and inference since the inception of MLPerf, including the MLPerf Inference 3.0 benchmarks released today.

“Three years ago when we introduced A100, the AI world was dominated by computer vision. Generative AI has arrived,” said NVIDIA founder and CEO Jensen Huang.

“This is exactly why we built Hopper, specifically optimized for GPT with the Transformer Engine. Today’s MLPerf 3.0 highlights Hopper delivering 4x more performance than A100.

“The next level of Generative AI requires new AI infrastructure to train large language models with great energy efficiency. Customers are ramping Hopper at scale, building AI infrastructure with tens of thousands of Hopper GPUs connected by NVIDIA NVLink and InfiniBand.

“The industry is working hard on new advances in safe and trustworthy Generative AI. Hopper is enabling this essential work,” he said.

The latest MLPerf results show NVIDIA taking AI inference to new levels of performance and efficiency from the cloud to the edge.

Specifically, NVIDIA H100 Tensor Core GPUs running in DGX H100 systems delivered the highest performance in every test of AI inference, the job of running neural networks in production. Thanks to software optimizations, the GPUs delivered up to 54% performance gains from their debut in September.

In healthcare, H100 GPUs delivered a 31% performance increase since September on 3D-UNet, the MLPerf benchmark for medical imaging.

Powered by its Transformer Engine, the H100 GPU, based on the Hopper architecture, excelled on BERT, a transformer-based large language model that paved the way for today’s broad use of generative AI.

Generative AI lets users quickly create text, images, 3D models and more. It’s a capability companies from startups to cloud service providers are rapidly adopting to enable new business models and accelerate existing ones.

Hundreds of millions of people are now using generative AI tools like ChatGPT — also a transformer model — expecting instant responses.

At this iPhone moment of AI, performance on inference is vital. Deep learning is now being deployed nearly everywhere, driving an insatiable need for inference performance from factory floors to online recommendation systems.

L4 GPUs Speed Out of the Gate

NVIDIA L4 Tensor Core GPUs made their debut in the MLPerf tests at over 3x the speed of prior-generation T4 GPUs. Packaged in a low-profile form factor, these accelerators are designed to deliver high throughput and low latency in almost any server.

L4 GPUs ran all MLPerf workloads. Thanks to their support for the key FP8 format, their results were particularly stunning on the performance-hungry BERT model.

In addition to stellar AI performance, L4 GPUs deliver up to 10x faster image decode, up to 3.2x faster video processing and over 4x faster graphics and real-time rendering performance.

Announced two weeks ago at GTC, these accelerators are already available from major systems makers and cloud service providers. L4 GPUs are the latest addition to NVIDIA’s portfolio of AI inference platforms launched at GTC.

Software, Networks Shine in System Test

NVIDIA’s full-stack AI platform showed its leadership in a new MLPerf test.

The so-called network-division benchmark streams data to a remote inference server. It reflects the popular scenario of enterprise users running AI jobs in the cloud with data stored behind corporate firewalls.

On BERT, remote NVIDIA DGX A100 systems delivered up to 96% of their maximum local performance, slowed in part because they needed to wait for CPUs to complete some tasks. On the ResNet-50 test for computer vision, handled solely by GPUs, they hit the full 100%.

Both results are thanks, in large part, to NVIDIA Quantum Infiniband networking, NVIDIA ConnectX SmartNICs and software such as NVIDIA GPUDirect.

Orin Shows 3.2x Gains at the Edge

Separately, the NVIDIA Jetson AGX Orin system-on-module delivered gains of up to 63% in energy efficiency and 81% in performance compared with its results a year ago. Jetson AGX Orin supplies inference when AI is needed in confined spaces at low power levels, including on systems powered by batteries.

For applications needing even smaller modules drawing less power, the Jetson Orin NX 16G shined in its debut in the benchmarks. It delivered up to 3.2x the performance of the prior-generation Jetson Xavier NX processor.

A Broad NVIDIA AI Ecosystem

The MLPerf results show NVIDIA AI is backed by the industry’s broadest ecosystem in machine learning.

Ten companies submitted results on the NVIDIA platform in this round. They came from the Microsoft Azure cloud service and system makers including ASUS, Dell Technologies, GIGABYTE, H3C, Lenovo, Nettrix, Supermicro and xFusion.

Their work shows users can get great performance with NVIDIA AI both in the cloud and in servers running in their own data centers.

NVIDIA partners participate in MLPerf because they know it’s a valuable tool for customers evaluating AI platforms and vendors. Results in the latest round demonstrate that the performance they deliver today will grow with the NVIDIA platform.

Users Need Versatile Performance

NVIDIA AI is the only platform to run all MLPerf inference workloads and scenarios in data center and edge computing. Its versatile performance and efficiency make users the real winners.

Real-world applications typically employ many neural networks of different kinds that often need to deliver answers in real time.

For example, an AI application may need to understand a user’s spoken request, classify an image, make a recommendation and then deliver a response as a spoken message in a human-sounding voice. Each step requires a different type of AI model.

The MLPerf benchmarks cover these and other popular AI workloads. That’s why the tests ensure IT decision makers will get performance that’s dependable and flexible to deploy.

Users can rely on MLPerf results to make informed buying decisions, because the tests are transparent and objective. The benchmarks enjoy backing from a broad group that includes Arm, Baidu, Facebook AI, Google, Harvard, Intel, Microsoft, Stanford and the University of Toronto.

Software You Can Use

The software layer of the NVIDIA AI platform, NVIDIA AI Enterprise,  ensures users get optimized performance from their infrastructure investments as well as the enterprise-grade support, security and reliability required to run AI in the corporate data center.

All the software used for these tests is available from the MLPerf repository, so anyone can get these world-class results.

Optimizations are continuously folded into containers available on NGC, NVIDIA’s catalog for GPU-accelerated software. The catalog hosts NVIDIA TensorRT, used by every submission in this round to optimize AI inference.

Read this technical blog for a deeper dive into the optimizations fueling NVIDIA’s MLPerf performance and efficiency.

NVIDIA today recognized a dozen partners in the Americas for their work enabling customers to build and deploy AI applications across a broad range of industries.

NVIDIA Partner Network (NPN) Americas Partner of the Year awards were given out to companies in 13 categories covering AI, consulting, distribution, education, healthcare, integration, networking, the public sector, rising star, service delivery, software and the Canadian market.  A new award category created this year recognizes growing AI adoption in retail, as leaders begin to introduce new AI-powered services addressing customer service, loss prevention and restocking analytics.

“NVIDIA’s commitment to driving innovation in AI has created new opportunities for partners to help customers leverage cutting-edge technology to reduce costs, grow opportunities and solve business challenges,” said Rob Enderle, president and principal analyst at the Enderle Group. “The winners of the 2023 NPN awards reflect a diverse group of AI business experts that have showcased deep knowledge in delivering transformative solutions to customers across a range of industries.”

The 2023 NPN award winners for the Americas are:

• Arrow Electronics — Distribution Partner of the Year. Recognized for providing end-to-end NVIDIA AI technologies across a variety of industries, such as manufacturing, retail, healthcare and robotics, to help organizations drive accelerated computing and robotics strategies via on-prem, hybrid cloud and intelligent edge solutions, and through Arrow’s Autonomous Machines Center of Excellence.
• Cambridge Computer — Higher Education Partner of the Year. Recognized for the third consecutive year for its continued focus on providing NVIDIA AI solutions to the education, life sciences and research computing sectors.
• CDW — Software Partner of the Year. Recognized for deploying NVIDIA AI and visualization solutions to customers from a broad range of industries and adopting deep industry expertise for end-to-end customer support.
• CDW Canada — Canadian Partner of the Year. Recognized for providing IT solutions that enable the nation’s leading vendors to offer customized solutions with NVIDIA technology, meeting the needs of each client.
• Deloitte — Consulting Partner of the Year. Recognized for the third consecutive year for creating new AI markets for clients by expanding AI investments in solutions developed with NVIDIA across enterprise AI, as well as expanding into new offerings with generative AI and NVIDIA DGX Cloud.
• FedData Technology Solutions — Rising Star Partner of the Year. Recognized for NVIDIA DGX-based design wins with key federal customers and emerging work with the NVIDIA Omniverse platform for building and operating metaverse applications.
• Insight — Retail Partner of the Year. Recognized for its deep understanding of the industry, ecosystem partnerships and the ability to orchestrate best-in-class solutions to bring real-time speed and predictability to retailers, enabling intelligent stores, intelligent quick-service restaurants, intelligent supply chain and omni-channel management.
• Lambda — Solution Integration Partner of the Year. Recognized for the third consecutive year for its commitment to providing end-to-end NVIDIA solutions, both on premises and in the cloud, across industries including higher education and research, the federal and public sectors, and healthcare and life sciences.
• Mark III — Healthcare Partner of the Year. Recognized for its unique team and deep understanding of the NVIDIA portfolio, which provides academic medical centers, research institutions, healthcare systems and life sciences organizations with NVIDIA infrastructure, software and cloud technologies to build out AI, HPC and simulation Centers of Excellence.
• Microway — Public Sector Partner of the Year. Recognized for its technical depth and engineering focus on servicing the public sector using technologies across the NVIDIA portfolio, including high performance computing and other specializations.
• Quantiphi — Service Delivery Partner of the Year. Recognized for the second consecutive year for its commitment to driving adoption of NVIDIA products in areas like generative AI services with customized large language models, digital avatars, edge computing, medical imaging and data science, as well as its expertise in helping customers build and deploy AI solutions at scale.
• World Wide Technology — AI Solution Provider of the  Year. Recognized for its leadership in driving adoption of the NVIDIA portfolio of AI and accelerated computing solutions, as well as its continued investments in AI infrastructure for large language models, computer vision, Omniverse-based digital twins, and customer testing and labs in the WWT Advanced Technology Center.
• World Wide Technology — Networking Partner of the Year. Recognized for its expertise driving NVIDIA high-performance networking solutions to support accelerated computing environments across multiple industries and AI solutions.

This year’s awards arrive as AI adoption is rapidly expanding across industries, unlocking new opportunities and accelerating discovery in healthcare, finance, business services and more. As AI models become more complex, the 2023 NPN Award winners are expert partners that can help enterprises develop and deploy AI in production using the infrastructure that best aligns with their operations.

Learn how to join the NPN, or find your local NPN partner.

Editor’s note: This post is part of our weekly In the NVIDIA Studio series, which celebrates featured artists, offers creative tips and tricks, and demonstrates how NVIDIA Studio technology improves creative workflows. We’re also deep diving on new GeForce RTX 40 Series GPU features, technologies and resources, and how they dramatically accelerate content creation.

AI-powered technology in creative apps, once considered nice to have, is quickly becoming essential for aspiring and experienced content creators.

Video editor Patrick Stirling used the Magic Mask feature in Blackmagic Design’s DaVinci Resolve software to create a custom effect that creates textured animations of people, this week In the NVIDIA Studio.

“I wanted to use ‘Magic Mask’ to replace subjects with textured, simplified, cut-out versions of themselves,” said the artist. “This style is reminiscent of construction-paper creations that viewers might have played with in childhood, keeping the energy of a scene while also pulling attention away from any specific features of the subject.”

Stirling’s original attempts to implement this effect were cut short due to the limitations of his six-year-old system. So Stirling built his first custom PC — equipped with a GeForce RTX 4080 GPU — to tackle the challenge. The difference was night and day, he said.

“I was able to find and maintain a creative flow so much more easily when I didn’t feel like I was constantly running into a wall and waiting for my system to catch up,” said Stirling.

The AI-powered Magic Mask feature, which allows quick selection of objects and people in a scene, was accelerated by his RTX 4080 GPU, delivering up to a 2x increase in AI performance over the previous generation. “The GPU also provides the power the DaVinci Neural Engine needs for some of these really cool effects,” said Stirling.

Stirling opened a short clip within the RTX GPU-accelerated Fusion page in DaVinci Resolve, a node-based workflow with hundreds of 2D and 3D tools. Nodes are popular as they make video editing a completely procedural process — allowing for non-linear, non-destructive workflows.

He viewed edits in real time using two windows opened side by side, with original footage on the left and node modifications on the right.

Stirling then drew blue lines to apply Magic Mask to each surface on the subject that he wanted to layer. As its name suggests, Magic Mask works like magic, but it’s not perfect. When the effect masked more than the extended jacket layer, Stirling drew a secondary red line to designate what not to capture in that area.

He applied similar techniques to the dress shirt, hands, beard, hair and facial skin. The artist then added generic colored backgrounds with Background nodes on each layer to complete his 2D character.

Stirling used Merge nodes to combine background and foreground images. He deployed the Fast Noise node to create two types of textures for the 2D man and the real-life footage, providing more contrast for the visual.

Stirling then added a color corrector to tweak saturation, his RTX GPU accelerating the process. He completed his video editing by combining the Magic Mask effect and all remaining nodes — Background, Merge and Fast Noise.

When it’s time to wrap up the project, Stirling can deploy the RTX 4080 GPU’s dual AV1 video encoders — which would cut export times in half.

Stirling encourages aspiring content creators to “stay curious” and “not ignore the value of connecting with other creative people.”

“Regularly being around people doing the same kind of work as you will constantly expose new methods and approaches for your own creative projects,” he said.

Check out Stirling’s YouTube channel for DaVinci Resolve tutorials.

Follow NVIDIA Studio on Instagram, Twitter and Facebook. Access tutorials on the Studio YouTube channel and get updates directly in your inbox by subscribing to the Studio newsletter.

It’s another rewarding GFN Thursday, with 23 new games for April on top of 11 joining the cloud this week and a new Marvel’s Midnight Suns reward now available first for GeForce NOW Premium members.

Newark, N.J., is next to complete its upgrade to RTX 4080 SuperPODs, making it the 12th region worldwide to bring new performance to Ultimate members.

GeForce NOW on SHIELD TV is being updated for a more consistent experience across Android and TV devices. Update 6.00 has begun rolling out to SHIELD TV owners this week.

Plus, work is underway to bring the initial batch of Xbox first-party games and features to GeForce NOW.

Teamwork Makes the Dream Work

Last month, we announced a partnership with Microsoft to bring Xbox Game Studios PC games to the GeForce NOW library, including titles from Bethesda, Mojang Studios and Activision, pending closure of Microsoft’s acquisition. It’s a shared commitment to giving gamers more choice and enabling PC gamers to play their favorite games anywhere.

Since then the teams at both companies have been collaborating on delivering a best-in-class cloud gaming experience that PC gamers have come to expect, delivering a seamless experience across any device, whether playing locally or in the cloud.

We’re making progress, and in future GFN Thursdays we will provide an update on onboarding of individual titles from Microsoft’s incredibly rich catalog of first party PC games. Stay tuned to our GFN Thursday updates for the latest.

Medieval Marvel

Starting today, Premium GeForce NOW members can claim their marvel-ous new reward. Marvel’s Midnight Suns, the tactical role-playing game from the creators of XCOM, has been praised for its immersive game play and cutting-edge visuals with support for DLSS 3 technology on top of RTX-powered ray tracing.

With the game’s first downloadable content, called The Good, The Bad, and The Undead, fans were thrilled to welcome Deadpool to the roster. This week, members can get their free reward to secure Captain Marvel’s Medieval Marvel suit.

Ultimate and Priority members can visit the GeForce NOW Rewards portal today and update the settings to start receiving special offers and in-game goodies. Better hurry, as this reward is available on a first-come, first-served basis only through Saturday, May 6.

April is FOOL of Games

No joke, kick the weekend off right by streaming Have a Nice Death. Restore order in this darkly charming 2D action game from Gearbox while playing as an overworked Death whose employees at Death Inc. have run rampant as caretakers of souls. Hack and slash through numerous minions and bosses in each department at the company, using unique weapons and spells.

This leads the 11 new games joining the cloud this week:

• 9 Years of Shadows (New release on Steam)
• Terra Nil (New release on Steam, March 28)
• Gripper (New release on Steam, March 29)
• Smalland: Survive the Wilds (New release on Steam, March 29)
• DREDGE (New release on Steam, March 30)
• Ravenbound (New release on Steam, March 30)
• The Great War: Western Front (New release on Steam, March 30)
• Troublemaker (New release on Steam, March 31)
• Have a Nice Death (Steam)
• Tower of Fantasy (Steam)
• Tunche (Free on Epic Games Store)

Plus, look forward to the rest of April:

• Meet Your Maker (New release on Steam, April 4)
• Road 96: Mile 0 (New release on Steam, April 4)
• TerraScape (New release on Steam, April 5)
• Curse of the Sea Rats (New release on Steam, April 6)
• Ravenswatch (New release on Steam, April 6)
• Supplice (New release on Steam, April 6)
• DE-EXIT – Eternal Matters (New release on Steam, April 14)
• Survival: Fountain of Youth (New release on Steam, April 19)
• Tin Hearts (New release on Steam, April 20)
• Dead Island 2 (New Release on Epic Games Store, April 21)
• Afterimage (New release on Steam, April 25)
• Roots of Pacha (New release on Steam, April 25)
• Bramble: The Mountain King (New release on Steam, April 27)
• 11-11 Memories Retold (Steam)
• canVERSE (Steam)
• Teardown (Steam)
• Get Even (Steam)
• Little Nightmares (Steam)
• Little Nightmares II (Steam)
• The Dark Pictures Anthology: Man of Medan (Steam)
• The Dark Pictures Anthology: Little Hope (Steam)
• The Dark Pictures Anthology: House of Ashes (Steam)
• The Dark Pictures Anthology: The Devil in Me (Steam)

More March Madness

On top of the 19 games announced in March, nine extra ones joined the GeForce NOW library this month, including this week’s additions 9 Years of Shadows, Terra Nil, Gripper, Troublemaker, Have a Nice Death, Tunche, as well as:

• Call of the Sea (Epic Games Store, March 9)
• GRID Legends (Steam and EA)
• Tchia (New release on Epic Games Store)

System Shock didn’t make it in March due to a shift in its release date, nor did Chess Ultra due to a technical issue.

With so many titles streaming from the cloud, what game will you play next? Let us know in the comments below, on Twitter or on Facebook.

What’s next on your list of games to try?

— NVIDIA GeForce NOW (@NVIDIAGFN) March 29, 2023

Editor’s note: This post is part of our weekly In the NVIDIA Studio series, which celebrates featured artists, offers creative tips and tricks, and demonstrates how NVIDIA Studio technology improves creative workflows. We’re also deep diving on new GeForce RTX 40 Series GPU features, technologies and resources, and how they dramatically accelerate content creation.

It’s a celebration, creators!

Blender, the world’s most popular 3D creation suite — free and open source — released its major version 3.5 update. Expected to have a profound impact on 3D creative workflows, this latest release features support for Open Shading Language (OSL) shaders with the NVIDIA OptiX ray-tracing engine.

Plus, 3D artist and filmmaker Pablo Reche Beltrán, aka AuraProds, joins the 50th edition of the In the NVIDIA Studio series this week to share his Jurassic Park-inspired short film. Thank you to the artists who’ve contributed to the series, those who influence and inspire each day and those who will do so in the future.

Finally, enter the Watch ‘n Learn Giveaway hosted by creative community 80LV for a chance to win a powerful GeForce RTX 4080 GPU. Enter by watching an Omniverse for creators GTC session, filling out this form, and tagging #GTC23 and @NVIDIAOmniverse on social media by Thursday, March 30.

Better Renders in Blender 3.5

Supporting OSL shaders with NVIDIA OptiX, the Blender update 3.5 now enables 3D artists to use shaders in Cycles and render them on an NVIDIA RTX GPU. Previously, such a workflow was limited to CPU rendering only.

This saves an extraordinary amount of time for artists, as scenes that use OSL can be completely rendered 36x faster than with a CPU alone.

Blender 3.5 also delivers updates to creative workflows in animation and rigging, nodes and physics, modeling and more.

Read the full release notes.

Dino-mite Renders Never Go Extinct

3D artist AuraProds fondly remembers watching the blockbuster movie Jurassic Park in theaters as a kid, wishing to one day recreate a memorable scene in which a giant T-Rex frightens the main characters who are huddled together in a car. Unlike that scary moment, however, the artist’s AuraProds video came together in a rather cute way.

“The concept art was made by my five- and eight-year-old cousins,” said AuraProds. “They drew the dinosaurs and inspired me to turn them into 3D.”

Based in Almería, a small city in southern Spain, AuraProds was perfectly situated to capture video footage in the nearby town of Tabernas, where Hollywood directors often shoot western movies. With the requisite footage captured, AuraProds modeled dinosaurs to populate the scene.

His preferred 3D app is Blender. “No doubt,” said AuraProds. ”I really like it because I can bring to 3D any idea in my head with a simple workflow.”

The artist modeled and sculpted each dinosaur by hand, using Blender Cycles RTX-accelerated OptiX ray tracing in the viewport for interactive, photorealistic modeling, thanks to his GeForce RTX 4080 GPU.

Satisfied with the models, AuraProds experimented with a variety of textures before moving on to rig and animate his models. Motion-blur visual effects were applied quickly with accelerated rendering and NVIDIA NanoVBD for easier rendering of volumes.

RTX-accelerated OptiX ray tracing in Blender Cycles allowed AuraProds to quickly export fully rendered files to Blackmagic Design’s DaVinci Resolve software.

Here, AuraProds RTX GPU dramatically accelerated his compositing workflows. GPU-accelerated color grading, video editing and color scopes were applied with ease. And decoding with NVDEC, also GPU accelerated, enabled buttery-smooth playback and scrubbing of high-resolution footage.

“NVIDIA RTX GPUs are the only technology that could handle the massive amount of polygons for this project. I know from my own experience how reliable the GPUs are.”’ — AuraProds.

The RTX 4080 GPU sped AuraProds use of new AI video editing tools, delivering up to a 2x increase in AI performance over the previous generation. Performance boosts were also applied to existing RTX-accelerated AI features — including automatic tagging of clips and tracking of effects, SpeedWarp for smooth slow motion and Video Super Resolution.

AuraProds then applied several AI features to achieve his desired visual effect. He wrapped up the project by deploying the RTX 4080 GPU’s dual AV1 video encoders — cutting the export time in half.

“I got to bring that aesthetic and memory to this new digital age with today’s AI tools,” said AuraProds.

For more of AuraProds video content, check out VELOX — a short 3D film shot entirely with a green screen — made by scanning an entire desert and creating several 3D spaceships over two months, available on his YouTube channel.

Experienced and aspiring content creators can discover exclusive step-by-step tutorials from industry-leading artists, inspiring community showcases and more on the NVIDIA Studio YouTube channel, which includes a curated Blender playlist.

Follow NVIDIA Studio on Instagram, Twitter and Facebook. Get updates directly in your inbox by subscribing to the Studio newsletter. 

Tools like ChatGPT have awakened the world to the potential of generative AI. Now, much more is coming.

On the latest episode of the NVIDIA AI Podcast, Yves Jacquier, executive director of Ubisoft La Forge, shares valuable insights into the transformative potential of generative AI in the gaming industry. With over two decades of experience in technology innovation, science and R&D management across various sectors, Jacquier’s comprehensive expertise makes him a true visionary in the field.

During his conversation with podcast host Noah Kravitz, Jacquier highlighted how generative AI, which enables computers to create unique content such as images, text and music, is already revolutionizing the gaming sector. By designing new levels, characters and items, and generating realistic graphics and soundscapes, this cutting-edge technology offers countless opportunities for more immersive and engaging experiences.

As the driving force behind Ubisoft La Forge, Jacquier plays a crucial role in shaping the company’s academic R&D strategy. Key milestones include establishing a chair in AI deep learning in 2011 and founding Ubisoft La Forge, the first lab in the gaming industry dedicated to applied academic research. This research is being translated into state-of-the-art gaming experiences.

Jacquier expressed confidence that generative AI will play a vital role in sculpting the gaming industry and providing unparalleled gaming experiences for enthusiasts around the world.

Related Articles

Sequoia Capital’s Pat Grady and Sonya Huang on Generative AI

Partners at Sequoia Capital, Pat Grady and Sonya Huang, discuss their thought-provoking essay, “Generative AI: A Creative New World.” The authors explore the potential of generative AI to unlock new realms of creativity and expression, while also addressing the challenges and ethical implications of this technology. They also provide insights into generative AI’s future.

Real or Not Real? Attorney Steven Frank Employs Deep Learning to Authenticate Art

Steven Frank, a partner at law firm Morgan Lewis, specializes in intellectual property and commercial technology law. He is part of a husband-wife duo that utilized convolutional neural networks to authenticate masterpieces, including da Vinci’s Salvador Mundi, with the aid of AI.

GANTheftAuto: Harrison Kinsley on AI-Crafted Gaming Environments

While humans playing games against machines is a familiar concept, computers can now develop games for people to enjoy. Programming aficionado and social media influencer Harrison Kinsley devised GANTheftAuto, an AI-driven neural network that produces a playable segment of the iconic video game Grand Theft Auto V.

Subscribe, Review and Follow NVIDIA AI on Twitter

If you found this episode insightful, subscribe to the NVIDIA AI Podcast on your preferred podcast platform and leave a rating and review. Stay connected with @NVIDIAAI on Twitter.

Gamers love games — as do the people who make them.

GeForce NOW streams over 1,500 games from the cloud, and with the Game Developers Conference in full swing this week, today’s GFN Thursday celebrates all things games: the tech behind them, the tools that bring them to the cloud, the ways to play them and the new ones being added to the library this week.

Developers use a host of NVIDIA resources to deliver the best in PC cloud gaming experiences. CD PROJEKT RED, one of many developers to tap into these resources, recently announced a new update coming to Cyberpunk 2077 on April 11 — including a new technology preview for Ray Tracing: Overdrive Mode that enables full ray tracing on GeForce RTX 40 Series GPUs and RTX 4080 SuperPODs.

In addition, members in and around Sofia, Bulgaria, can now experience the best of GeForce NOW Ultimate cloud gaming. It’s the latest city to roll out RTX 4080 gaming rigs to GeForce NOW servers around the globe.

Plus, with five new games joining the cloud this week, and an upcoming marvel-ous reward, GeForce NOW members can look forward to a busy weekend of streaming goodness.

Developers, Mount Up!

GDC presents the ideal time to spotlight GeForce NOW tools that enable developers to seamlessly bring their games to the cloud. NVIDIA tools, software development kits (SDKs) and partner engines together enable the production of stunning real-time content that uses AI and ray tracing. And bringing these games to billions of non-PC devices is as simple as checking an opt-in box.

GeForce NOW taps into existing game stores, allowing game developers to reap the benefits of a rapidly growing audience without the hassle of developing for another platform. This means zero port work to bring games to the cloud. Users don’t have to buy games for another platform and can play them on many of the devices they already own.

Developers who want to do more have access to the GeForce NOW Developer Platform — an SDK and toolset empowering integration of, interaction with and testing on the NVIDIA cloud gaming service. It allows developers to enhance their games to run more seamlessly, add cloud gaming into their stores and launchers, and let users connect their accounts and libraries to GeForce NOW.

The SDK is a set of APIs, runtimes, samples and documentation that allows games to query for cloud execution and enable virtual touchscreens; launchers to trigger cloud streaming of a specified game; and GeForce NOW and publisher backends to facilitate account linking and game library ownership syncing, already available for Steam and Ubisoft games.

Content developers have a slew of opportunities to bring their virtual worlds and interactive experiences to users in unique ways, powered by the cloud.

Metaverse services company Improbable will use NVIDIA cloud gaming infrastructure for an interactive, live, invite-only experience that will accommodate up to 10,000 guests. Other recent developer events included the DAF Trucks virtual experience, where potential customers took the newest DAF truck for a test drive in a simulated world, with PixelMob’s Euro Truck Simulator 2 providing the virtual playground.

Furthermore, CD PROJEKT RED will be delivering full ray tracing, aka path tracing, to Cyberpunk 2077. Such effects were previously only possible for film and TV. With the power of a GeForce RTX 4080 gaming rig in the cloud, Ultimate members will be able to stream the new technology preview for the Ray Tracing: Overdrive Mode coming to Cyberpunk 2077 across devices — even Macs — no matter the game’s system requirements.

Get Ready for a Marvel-ous Week

GeForce NOW Ultimate members have been enjoying Marvel’s Midnight Suns’ ultra-smooth, cinematic game play thanks to DLSS 3 technology support on top of RTX-powered ray tracing, which together enable graphics breakthroughs.

Now, members can fight among the legends with Captain Marvel’s Medieval Marvel suit in a free reward, which will become available at the end of the month — first to Premium members who are opted into GeForce NOW rewards. This reward is only available until May 6, so upgrade to an Ultimate or Priority membership today and opt into rewards to get first access.

Next, on to the five new games hitting GeForce NOW week for a happy weekend:

• Tchia (New release on Epic Games Store)
• Chess Ultra (New release on Epic Games Store, March 23)
• Amberial Dreams (Steam)
• Symphony of War: The Nephilim Saga (Steam)
• No One Survived (Steam)

And with that, we’ve got a question to end this GFN Thursday:

You’ve got free rent for a year to live in a video game city of your choice, which one are you choosing?

— NVIDIA GeForce NOW (@NVIDIAGFN) March 22, 2023

Game developer CD PROJEKT RED today at the Game Developers Conference in San Francisco unveiled a technology preview for Cyberpunk 2077 with path tracing, coming April 11.

Path tracing, also known as full ray tracing, accurately simulates light throughout an entire scene. It’s used by visual effects artists to create film and TV graphics that are indistinguishable from reality. But until the arrival of GeForce RTX GPUs with RT Cores, and the AI-powered acceleration of NVIDIA DLSS, real-time video game path tracing was impossible because it is extremely GPU intensive.

“This not only gives better visuals to the players but also has the promise to revolutionize the entire pipeline of how games are being created,” said Pawel Kozlowski, a senior technology developer engineer at NVIDIA.

This technology preview, Cyberpunk 2077’s Ray Tracing: Overdrive Mode, is a sneak peak into the future of full ray tracing. With full ray tracing, now practically all light sources cast physically correct soft shadows. Natural colored lighting also bounces multiple times throughout Cyberpunk 2077’s world, creating more realistic indirect lighting and occlusion.

Cyberpunk 2077, previously an early adopter of ray tracing, becomes the latest modern blockbuster title to harness real-time path tracing. Coming shortly after path tracing for Minecraft, Portal and Quake II, it underscores a wave of adoption in motion.

Like with ray tracing, it’s expected many more will follow. And the influence on video games is just the start, as real-time path tracing holds promise for many design industries.

Decades of Research Uncorked

Decades in the making, real-time path tracing is indeed a big leap in gaming graphics.

While long used in computer-generated imagery for movies, path tracing there took place in offline rendering farms, often requiring hours to render a single frame.

In gaming, which requires fast frame rates, rendering needs to happen in about 0.016 seconds.

Since the 1970s, video games have relied on rasterization techniques (see below). More recently, in 2018, NVIDIA introduced RTX GPUs to support ray tracing. Path tracing is the final frontier for the most physically accurate lighting and shadows.

Path tracing has been one of the main lighting algorithms used in offline rendering farms and computer graphics in films for years. It wasn’t until GeForce RTX 40 series and DLSS 3 was available that it was possible to bring path tracing to real-time graphics.

Cyberpunk 2077 also taps into Shader Execution Reordering — available for use on the NVIDIA Ada Lovelace architecture generation — which optimizes GPU workloads, enabling more efficient path-traced lighting.

Accelerated by DLSS 3  

DLSS 3 complements groundbreaking advancements in path tracing and harnesses modern AI — built on GPU-accelerated deep learning, a form of neural networking — as a powerful gaming performance multiplier. DLSS allows games to render 1/8th of the pixels, then uses AI and GeForce RTX Tensor Cores to reconstruct the rest, dramatically multiplying frame rates, while delivering crisp, high-quality images that rival native resolution.

Running on Ada Lovelace advances — launched with GeForce RTX 40 Series GPUs — DLSS 3 multiplies frame rates, maintaining image quality and responsiveness in games.

Powerful Tools Now Available 

For game developers, NVIDIA today at GDC announced the availability of the RTX+ Path Tracing SDK 1.0. The package of technologies includes DLSS 3, Shader Execution Reordering (SER), RTX Direct Illumination (RTXDI) and NVIDIA Real-Time Denoisers (NRD).

Learn more about full RTX path tracing. 

And catch up on all the breakthroughs in generative AI and the metaverse by joining us at GTC this week.

Gamers wanted better graphics. GPUs delivered. Those GPUs became the key to the world-changing AI revolution. Now gamers are reaping the benefits.

At GDC 2023 in San Francisco this week, the gaming industry’s premier developers conference, NVIDIA made a series of announcements, including new games and game development tools that promise to accelerate innovations at the intersection of neural networking and graphics, or neural graphics revolution.

DLSS 3 harnesses modern AI — built on GPU-accelerated deep learning, a form of neural networking — as a powerful gaming performance multiplier. DLSS allows games to render 1/8th of the pixels, then uses AI and GeForce RTX Tensor Cores to reconstruct the rest, dramatically multiplying frame rates, while delivering crisp, high-quality images that rival native resolution.

It’s just one example of how gamers are benefiting from the advancements in AI supercomputing showcased at this week’s NVIDIA GTC technology conference, which is running concurrently with GDC. And game developers are adopting DLSS at a break-neck pace.

DLSS complements ground-breaking advancements in ray tracing — a technology long used by filmmakers — to bring richer, more immersive visual experiences to gamers in real time.

Thanks, in part, to DLSS, real-time ray tracing, considered by many to be impossible in 2017, exploded onto the gaming scene with the debut of NVIDIA RTX in 2018. Ray tracing is now everywhere in games.

At GDC, game developer CD PROJEKT RED announced Cyberpunk 2077 will activate full ray tracing, also called path tracing, with the upcoming technology preview of Ray Tracing: Overdrive Mode in Cyberpunk 2077.

Cyberpunk 2077, previously an early adopter of ray tracing, becomes the latest modern blockbuster title to harness real-time path tracing following  Minecraft, Portal and Quake II.

New DLSS 3 AAA Games Coming

During GDC, NVIDIA also announced the addition of DLSS 3 support for even more popular AAA games, including Diablo IV, Forza Horizon 5 and Redfall. 

• Diablo IV, the latest installment of the genre-defining Diablo franchise with multiple games, will launch on June 6 with DLSS 3, with ray tracing coming post-launch.
• Forza Horizon 5, named the best open-world racing game of all time, will update to DLSS 3 on March 28. 
• Redfall, Bethesda’s highly anticipated, open-world, co-op first-person shooter from Arkane Austin is launching on May 2 with DLSS 3, with ray tracing coming post-launch.

DLSS is now available in 270+ games and apps, and DLSS 3 is multiplying performance in 28 released games and has been adopted 7x faster than DLSS 2 in the first six months of their respective launches.

DLSS Frame Generation Now Publicly Available for Developers

NVIDIA announced DLSS Frame Generation is now publicly available for developers to integrate into their games and applications.

The public release of DLSS Frame Generation plug-ins will allow even more developers to adopt the framerate-boosting technology.

DLSS Frame Generation is now available via NVIDIA Streamline, an open-source, cross-vendor framework that simplifies the integration of super-resolution technologies in 3D games and apps.

For all the details, dig into our full coverage on GeForce News.

Unreal Engine 5.2 Integration to Speed Up DLSS 3 Adoption

At GDC, NVIDIA and Epic announced the integration of DLSS 3 into the popular Unreal Engine game engine.

Unreal Engine is an open and advanced real-time 3D creation tool that gives game developers and creators the freedom and control to deliver cutting-edge 3D content, interactive experiences and immersive virtual worlds.

A DLSS 3 plug-in will debut in UE 5.2, making it more straightforward for any developer to accelerate the performance of their games and applications, further accelerating the adoption of DLSS.

Cyberpunk 2077: A Showcase for What’s Next

CD PROJEKT RED showcases a technology preview of path tracing with Cyberpunk 2077.

Path tracing, also known as full ray tracing, allows developers to create cinematic experiences. Simulating the physics of light, using ray tracing as part of a neural graphics system, it’s capable of photorealism in 3D settings for more dynamic lighting and shadows.

GeForce gamers will be able to activate full ray tracing with the upcoming technology preview of Ray Tracing: Overdrive Mode on April 11.

These advancements are anchored in NVIDIA RTX technologies. To bring these incredible effects to life, CD PROJEKT RED and NVIDIA have worked hand in hand to integrate NVIDIA DLSS 3 and introduce new optimizations for this entirely new, fully ray-traced pipeline.

NVIDIA Shader Execution Reordering helps GPUs execute incoherent workloads boosting performance; NVIDIA Real-Time Denoisers have been used to improve performance and image quality.

As a result, with full ray tracing, now practically all light sources cast physically correct soft shadows. Natural colored lighting also bounces multiple times throughout Cyberpunk 2077’s world, creating more realistic indirect lighting and occlusion.

More to Come

Cyberpunk 2077 is a case study of how GPUs have unlocked the AI revolution and will bring great experiences to PC gamers for years to come.

An expanding game roster, game engine support and continued improvements in performance and image quality are securing NVIDIA DLSS as a landmark technology for the neural graphics revolution in PC gaming.

Like old friends catching up over coffee, two industry icons reflected on how modern AI got its start, where it’s at today and where it needs to go next.

Jensen Huang, founder and CEO of NVIDIA, interviewed AI pioneer Ilya Sutskever in a fireside chat at GTC. The talk was recorded a day after the launch of GPT-4, the most powerful AI model to date from OpenAI, the research company Sutskever co-founded.

They talked at length about GPT-4 and its forerunners, including ChatGPT. That generative AI model, though only a few months old, is already the most popular computer application in history.

Their conversation touched on the capabilities, limits and inner workings of the deep neural networks that are capturing the imaginations of hundreds of millions of users.

Compared to ChatGPT, GPT-4 marks a “pretty substantial improvement across many dimensions,” said Sutskever, noting the new model can read images as well as text.

“In some future version, [users] might get a diagram back” in response to a query, he said.

Under the Hood With GPT

“There’s a misunderstanding that ChatGPT is one large language model, but there’s a system around it,” said Huang.

In a sign of that complexity, Sutskever said OpenAI uses two levels of training.

The first stage focuses on accurately predicting the next word in a series. Here, “what the neural net learns is some representation of the process that produced the text, and that’s a projection of the world,” he said.

The second “is where we communicate to the neural network what we want, including guardrails … so it becomes more reliable and precise,” he added.

Present at the Creation

While he’s at the swirling center of modern AI today, Sutskever was also present at its creation.

In 2012, he was among the first to show the power of deep neural networks trained on massive datasets. In an academic contest, the AlexNet model he demonstrated with AI pioneers Geoff Hinton and Alex Krizhevsky recognized images faster than a human could.

Huang referred to their work as the Big Bang of AI.

The results “broke the record by such a large margin, it was clear there was a discontinuity here,” Huang said.

The Power of Parallel Processing

Part of that breakthrough came from the parallel processing the team applied to its model with GPUs.

“The ImageNet dataset and a convolutional neural network were a great fit for GPUs that made it unbelievably fast to train something unprecedented,” Sutskever said.

That early work ran on a few GeForce GTX 5080 GPUs in a University of Toronto lab. Today, tens of thousands of the latest NVIDIA A100 and H100 Tensor Core GPUs in the Microsoft Azure cloud service handle training and inference on models like ChatGPT.

“In the 10 years we’ve known each other, the models you’ve trained [have grown by] about a million times,” Huang said. “No one in computer science would have believed the computation done in that time would be a million times larger.”

“I had a very strong belief that bigger is better, and a goal at OpenAI was to scale,” said Sutskever.

A Billion Words

Along the way, the two shared a laugh.

“Humans hear a billion words in a lifetime,” Sutskever said.

“Does that include the words in my own head,” Huang shot back.

“Make it 2 billion,” Sutskever deadpanned.

The Future of AI

They ended their nearly hour-long talk discussing the outlook for AI.

Asked if GPT-4 has reasoning capabilities, Sutskever suggested the term is hard to define and the capability may still be on the horizon.

“We’ll keep seeing systems that astound us with what they can do,” he said. “The frontier is in reliability, getting to a point where we can trust what it can do, and that if it doesn’t know something, it says so,” he added.

“Your body of work is incredible … truly remarkable,” said Huang in closing the session. “This has been one of the best beyond Ph.D. descriptions of the state of the art of large language models,” he said.

To get all the news from GTC, watch the keynote below.